Activity Monitor.
Activity Monitor and vial.
Plot:
Model:
hour.mod <- lm(Hourly_activity ~ Predation + Predation:Population + hour + monitor + start_day,data=dat.hourly_2)
summary(hour.mod)
# Accounting for the auto-correlation
gls.mod <- gls(Hourly_activity ~ Predation + Predation:Population + hour + monitor + start_day,
correlation = corAR1(form =~hour|individual),
data=dat.hourly_2)
anova(gls.mod)
summary(gls.mod)
gls.mod.2 <- gls(Hourly_activity ~ Predation + Predation:Population + light + light:Predation + hour + monitor + start_day,
correlation = corAR1(form =~hour|individual),
data=dat.hourly_2)
summary(gls.mod.2)
## Generalized least squares fit by REML
## Model: Hourly_activity ~ Predation + Predation:Population + light + light:Predation + hour + monitor + start_day
## Data: dat.hourly_2
## AIC BIC logLik
## 65642.46 65735.65 -32807.23
##
## Correlation Structure: AR(1)
## Formula: ~hour | individual
## Parameter estimate(s):
## Phi
## 0.6782841
##
## Coefficients:
## Value Std.Error t-value p-value
## (Intercept) 241.34462 23.068553 10.462061 0.0000
## PredationMantids -49.09869 16.540037 -2.968475 0.0030
## PredationSpider -54.05986 16.540037 -3.268424 0.0011
## lightlight -2.40636 5.614807 -0.428574 0.6682
## hour 0.46920 0.343496 1.365948 0.1720
## monitor2 -20.42689 5.399740 -3.782940 0.0002
## start_day -4.67585 0.827698 -5.649219 0.0000
## PredationControl:Population -5.05300 4.182621 -1.208094 0.2271
## PredationMantids:Population -1.64032 4.182621 -0.392176 0.6949
## PredationSpider:Population 3.00350 4.182621 0.718091 0.4727
## PredationMantids:lightlight 22.49193 7.854511 2.863569 0.0042
## PredationSpider:lightlight 29.77666 7.854511 3.791027 0.0002
##
## Correlation:
## (Intr) PrdtnM PrdtnS lghtlg hour montr2
## PredationMantids -0.358
## PredationSpider -0.358 0.500
## lightlight -0.078 0.141 0.141
## hour -0.156 0.000 0.000 -0.147
## monitor2 -0.117 0.000 0.000 0.000 0.000
## start_day -0.840 0.000 0.000 0.000 0.000 0.000
## PredationControl:Population -0.453 0.632 0.632 0.000 0.000 0.000
## PredationMantids:Population 0.000 -0.632 0.000 0.000 0.000 0.000
## PredationSpider:Population 0.000 0.000 -0.632 0.000 0.000 0.000
## PredationMantids:lightlight 0.072 -0.202 -0.101 -0.699 0.000 0.000
## PredationSpider:lightlight 0.072 -0.101 -0.202 -0.699 0.000 0.000
## strt_d PrdC:P PrdM:P PrdS:P PrdtM:
## PredationMantids
## PredationSpider
## lightlight
## hour
## monitor2
## start_day
## PredationControl:Population 0.000
## PredationMantids:Population 0.000 0.000
## PredationSpider:Population 0.000 0.000 0.000
## PredationMantids:lightlight 0.000 0.000 0.000 0.000
## PredationSpider:lightlight 0.000 0.000 0.000 0.000 0.500
##
## Standardized residuals:
## Min Q1 Med Q3 Max
## -1.5228457 -0.7029253 -0.2660428 0.4731306 10.5374245
##
## Residual standard error: 97.27581
## Degrees of freedom: 5760 total; 5748 residual
What are we trying to achieve? We want to fit a model that looks something like this
new.model <- glmer(Hourly_activity ~ Predation + Predation:Population + light + light:Predation + **f(hour)** + monitor + start_day
+ (1 + **f(hour)** + light | individual),
correlation = corAR1(form =~hour|individual))
# Do we still need this for residual variance?).
The question is, what is the functional form of *f()`?
man.mod <- lm(activity_counts ~ Treatment + hour + monitor,data=Mantid_hour)
summary(man.mod)
#pacf(resid(man.mod))
man_mod_2 <- gls(activity_counts ~ Treatment + hour + monitor, correlation = corAR1(form = ~ 1|hour), data=Mantid_hour)
anova(man_mod_2)
summary(man_mod_2)
#acf(resid(man_mod_2))
man_mod_3 <- gls(activity_counts ~ Treatment + light + light:Treatment + hour + monitor, correlation = corAR1(form =~1|hour), control = list(singular.ok = TRUE), data=Mantid_hour)
summary(man_mod_3)
## Generalized least squares fit by REML
## Model: activity_counts ~ Treatment + light + light:Treatment + hour + monitor
## Data: Mantid_hour
## AIC BIC logLik
## 7576.585 7614.001 -3780.292
##
## Correlation Structure: AR(1)
## Formula: ~1 | hour
## Parameter estimate(s):
## Phi
## -0.02507329
##
## Coefficients:
## Value Std.Error t-value p-value
## (Intercept) 15.377915 2.518158 6.106810 0.0000
## TreatmentMantid 3.022189 2.757262 1.096083 0.2734
## lightlight 12.207182 3.010944 4.054270 0.0001
## hour -0.299871 0.171128 -1.752317 0.0801
## monitor2 -1.367636 1.932598 -0.707667 0.4794
## TreatmentMantid:lightlight -9.470931 3.823295 -2.477164 0.0135
##
## Correlation:
## (Intr) TrtmnM lghtlg hour montr2
## TreatmentMantid -0.546
## lightlight -0.277 0.458
## hour -0.506 0.000 -0.442
## monitor2 -0.376 -0.003 0.005 -0.011
## TreatmentMantid:lightlight 0.395 -0.721 -0.635 0.000 0.000
##
## Standardized residuals:
## Min Q1 Med Q3 Max
## -0.8787072 -0.5620015 -0.3474958 0.2376949 11.2351921
##
## Residual standard error: 27.63892
## Degrees of freedom: 800 total; 794 residual
#confint(man_mod_3)
#acf(resid(man_mod_3))
Plot:
Model:
hour.mod <- lm(activity_counts ~ Treatment + hour + monitor,data=act_hour)
summary(hour.mod)
#pacf(resid(hour.mod))
correl_mod <- gls(activity_counts ~ Treatment + hour + monitor, correlation = corAR1(form = ~ 1|hour), data=act_hour)
anova(correl_mod)
summary(correl_mod)
#acf(resid(correl_mod))
act_cor_light_mod <- gls(activity_counts ~ Treatment + light + light:Treatment + hour + monitor, correlation = corAR1(form =~1|hour), control = list(singular.ok = TRUE), data=act_hour)
summary(act_cor_light_mod)
## Generalized least squares fit by REML
## Model: activity_counts ~ Treatment + light + light:Treatment + hour + monitor
## Data: act_hour
## AIC BIC logLik
## 8740.556 8777.972 -4362.278
##
## Correlation Structure: AR(1)
## Formula: ~1 | hour
## Parameter estimate(s):
## Phi
## 0.189614
##
## Coefficients:
## Value Std.Error t-value p-value
## (Intercept) 58.46631 6.404279 9.129258 0.0000
## TreatmentSpider -16.58233 7.075310 -2.343690 0.0193
## lightlight 17.27978 7.807945 2.213102 0.0272
## hour -1.21836 0.447145 -2.724753 0.0066
## monitor2 -14.73985 4.365534 -3.376414 0.0008
## TreatmentSpider:lightlight -5.28255 9.816117 -0.538151 0.5906
##
## Correlation:
## (Intr) TrtmnS lghtlg hour montr2
## TreatmentSpider -0.542
## lightlight -0.277 0.453
## hour -0.524 0.000 -0.443
## monitor2 -0.324 -0.030 -0.001 0.000
## TreatmentSpider:lightlight 0.397 -0.720 -0.629 0.000 0.002
##
## Standardized residuals:
## Min Q1 Med Q3 Max
## -1.0426789 -0.5293192 -0.2776230 0.1330279 8.8264125
##
## Residual standard error: 58.62376
## Degrees of freedom: 800 total; 794 residual
#confint(act_cor_light_mod)
#acf(resid(act_cor_light_mod))
Exp2_mod <- lm(activity_counts ~ Treatment + hour + monitor,data=Exp2_hour)
summary(Exp2_mod)
#pacf(resid(Exp2_mod))
Exp2_mod_2 <- gls(activity_counts ~ Treatment + hour + monitor, correlation = corAR1(form = ~ 1|hour), data=Exp2_hour)
anova(Exp2_mod_2)
summary(Exp2_mod_2)
#acf(resid(Exp2_mod_2))
Exp2_mod_3 <- gls(activity_counts ~ Treatment + light + light:Treatment + hour + monitor, correlation = corAR1(form =~1|hour), control = list(singular.ok = TRUE), data=Exp2_hour)
summary(Exp2_mod_3)
## Generalized least squares fit by REML
## Model: activity_counts ~ Treatment + light + light:Treatment + hour + monitor
## Data: Exp2_hour
## AIC BIC logLik
## 9280.294 9318.027 -4632.147
##
## Correlation Structure: AR(1)
## Formula: ~1 | hour
## Parameter estimate(s):
## Phi
## 0.3532345
##
## Coefficients:
## Value Std.Error t-value p-value
## (Intercept) 69.48297 8.797518 7.898020 0.0000
## TreatmentSpider 18.71113 9.541616 1.961003 0.0502
## lightlight 19.23095 10.557260 1.821585 0.0689
## hour -2.55429 0.620627 -4.115655 0.0000
## monitor2 -20.94594 5.445182 -3.846693 0.0001
## TreatmentSpider:lightlight 7.96490 13.056399 0.610038 0.5420
##
## Correlation:
## (Intr) TrtmnS lghtlg hour montr2
## TreatmentSpider -0.542
## lightlight -0.280 0.452
## hour -0.529 0.000 -0.446
## monitor2 -0.319 0.000 0.002 -0.001
## TreatmentSpider:lightlight 0.396 -0.731 -0.618 0.000 0.000
##
## Standardized residuals:
## Min Q1 Med Q3 Max
## -1.2782087 -0.6403775 -0.2858197 0.2660999 4.7367746
##
## Residual standard error: 68.9982
## Degrees of freedom: 832 total; 826 residual
#confint(Exp2_mod_3)
#acf(resid(Exp2_mod_3))
exp3.mod <- lm(activity_counts ~ Treatment + hour + monitor,data=Exp3_hour)
summary(exp3.mod)
#pacf(resid(exp3.mod))
exp3.mod_2 <- gls(activity_counts ~ Treatment + hour + monitor, correlation = corAR1(form = ~ 1|hour), data=Exp3_hour)
anova(exp3.mod_2)
summary(exp3.mod_2)
#acf(resid(exp3.mod_2))
exp3.mod_3 <- gls(activity_counts ~ Treatment + light + light:Treatment + hour + monitor, correlation = corAR1(form =~1|hour), control = list(singular.ok = TRUE), data=Exp3_hour)
summary(exp3.mod_3)
## Generalized least squares fit by REML
## Model: activity_counts ~ Treatment + light + light:Treatment + hour + monitor
## Data: Exp3_hour
## AIC BIC logLik
## 17824.65 17888.64 -8900.325
##
## Correlation Structure: AR(1)
## Formula: ~1 | hour
## Parameter estimate(s):
## Phi
## 0.3474537
##
## Coefficients:
## Value Std.Error t-value p-value
## (Intercept) 114.69176 11.256779 10.188684 0.0000
## TreatmentF -39.68239 12.971874 -3.059110 0.0023
## TreatmentSC -24.71865 13.170769 -1.876781 0.0607
## TreatmentSF -24.24473 13.392779 -1.810283 0.0704
## lightlight -13.79016 14.300861 -0.964289 0.3351
## hour -2.99360 0.576370 -5.193894 0.0000
## monitor2 6.77289 4.997551 1.355243 0.1755
## TreatmentF:lightlight 22.52772 17.521953 1.285685 0.1987
## TreatmentSC:lightlight 28.50192 17.749811 1.605759 0.1085
## TreatmentSF:lightlight 41.87207 18.042923 2.320692 0.0204
##
## Correlation:
## (Intr) TrtmnF TrtmSC TrtmSF lghtlg hour montr2
## TreatmentF -0.663
## TreatmentSC -0.690 0.588
## TreatmentSF -0.680 0.560 0.599
## lightlight -0.519 0.524 0.542 0.534
## hour -0.384 0.000 0.000 0.000 -0.303
## monitor2 -0.214 -0.013 0.011 0.012 0.002 0.000
## TreatmentF:lightlight 0.492 -0.740 -0.435 -0.415 -0.708 0.001 0.001
## TreatmentSC:lightlight 0.511 -0.436 -0.742 -0.444 -0.730 0.001 -0.004
## TreatmentSF:lightlight 0.504 -0.416 -0.444 -0.742 -0.720 0.000 -0.002
## TrtmF: TrtSC:
## TreatmentF
## TreatmentSC
## TreatmentSF
## lightlight
## hour
## monitor2
## TreatmentF:lightlight
## TreatmentSC:lightlight 0.588
## TreatmentSF:lightlight 0.562 0.597
##
## Standardized residuals:
## Min Q1 Med Q3 Max
## -1.4204762 -0.6420698 -0.3261980 0.2820635 5.0138722
##
## Residual standard error: 85.50982
## Degrees of freedom: 1539 total; 1529 residual
#confint(exp3.mod_3)
#acf(resid(exp3.mod_3))